CivArchive
    Vision Vanguard 17 (Krea 2) - V2026.1
    NSFW
    Preview 135824846
    Preview 135824868
    Preview 135824906
    Preview 135825161
    Preview 135824879
    Preview 135824872
    Preview 135825527
    Preview 135824881
    Preview 135825338
    Preview 135825014
    Preview 135824914
    Preview 135824887
    Preview 135824891
    Preview 135824897
    Preview 135824899
    Preview 135824905
    Preview 135824989
    Preview 135825083
    Preview 135825224
    Preview 135825364

    Vision Vanguard 17 [Krea 2]

    Raw cinematic realism with deep textured shadows, hard directional light and organic film grain. Trained on 1,200+ photos.

    • Base model: Krea 2

    • Trigger words: none required. Opening the prompt with "A photograph captures…" anchors the style.

    • Helpful keywords: high contrast, sharp focus, shallow depth of field, harsh lighting, gritty texture, cinematic realism, somber atmosphere, dramatic chiaroscuro, nocturnal, muted tones

    • Strength: 0.8–1.0. Use 0.9–1.0 for the strongest grit and contrast. The version specs list 0.85–1.0.

    • Negative prompt (optional): flat lighting, low contrast, saturated colors, blurry subject

    Prompt pattern: [Photograph type] captures [subject] [action] in [environment]. [Lighting] creates [atmosphere]. [Technical details].

    Example: A documentary photograph captures a weathered man staring into a storm on a rocky coastline. Harsh, cool moonlight creates a tense, mysterious atmosphere. Sharp focus on facial texture, deep shadows, and visible film grain.

    Tips

    • Black backgrounds and deep shadows work well.

    • Name materials and light sources specifically.

    • Use "minimalist" with lots of negative space for silhouettes.

    • Use "macro" or "underwater macro" for small subjects.

    Description

    • Base model: Krea 2

    • Strength: 0.85–1.0

    • Resolution: trained at 512, generate at 1024

    • CFG: 2.5–4.5 (Base) / 1.0 (Turbo)

    • Steps: 35–50 (Base) / 8–9 (Turbo)

    FAQ

    Comments (2)

    Melodias102Jul 6, 2026
    CivitAI

    Love the Vanguard LoRa series. Really adds some sweet dark depth to the images.

    thedivergentai
    Author
    Jul 7, 2026· 1 reaction

    Yay yay yay. I'm so happy. In contrast the dark deep images.

    I'm really glad to hear you like the LoRAs. I've been self-studying training for about 2 years now. So I really hope that LoRAs I've been putting out are nice and HQ.

    The VV series is trained on generally deep emotive images. So they'll all be great at telling stories.

    I hope people use them to generate some really cool stuff. I want to see like the realism applied to unexpected stuff. pretty sure it'd make some cool like aliens and beings.

    Anyways, thanks for the comment and enjoy generating.

    LORA
    Krea 2

    Details

    Downloads
    409
    Platform
    CivitAI
    Platform Status
    Available
    Created
    7/5/2026
    Updated
    11/20/2026
    Deleted
    -

    Files

    (Krea 2) Vision Vanguard 17 V2026.1.safetensors